Hey folks,
I am looking to create a new article to supplement the one that I created before on setting up CUDA with fedora.
Currently, the article I have in mind is an updated one of the above but with even more BONUS sections including:
- desktop setups including: setting up gnome tweaks, hardware acceleration on firefox, and so one
- data science notebooks: setting up a jupyterlab and marimo to use with CUDA 13 including using uv for keeping everything together and latest pytorch with an example running
- Going through a C++ thrust example (maybe writing a mandlebrot version in a comment)
- setting up a local ai stack (NVFP4/MXFP4 llm, visual language llm, RAG, websearch, openwebui chatgpt alternative and tools calling and integration with local text editors and IDE)
Is this too much? What do folks want to see the most? Is there something folks get stuck on?
I want to do this for the people who know what they are doing. Think senior developers, devops, SRE, data scientists who know their way around linux and want to get things working from scratch locally. Article is ready-ish but I want to add more (or trim it down) based on feedback before posting.
EDIT: I want to also gear this article to those that have a single blackwell GPU with 32GB-96GB VRAM.